Drug indicators such as the average number of drugs per prescription, antibiotic prescription rates, injection rates, average cost of prescription and drug-specific prescription rates (in this case, chloroquine and quinine), are significant indicators of appropriateness of prescribing. And while the indicators are important, what are more important are those factors that significantly effect a change in the indicators. Training and supervision previously mentioned in the literature and substantiated here in the authors' research, positively influenced the average number of drugs per prescription (Guiscafre et al 1988, Angunawela, Diwant & Tomson 1991, Guiterrez et al 1994, Chowdhury et al 1995, de Vries et al 1995, Gani, Tangkilisan & Pujilestary 1995, Sunartono & Darminto 1995, Chowdhury et al 1996). In facilities where the majority of prescribers had received some formal training in diagnosis and prescription, specifically training of at least six weeks, the average number of drugs prescribed was significantly lower. In facilities where prescribers received on-site or monthly doctor's supervision the average number of drugs prescribed was lower. In addition, in facilities where average prescribing experience was more than nine years and where at least half of the screeners had received high school or university education, the average number of drugs prescribed was also significantly lower.

Periodic modifications in the length of the screening course at the CBC Training School provided an opportunity to compare the average number of drug per prescription rates between facilities staffed by personnel attending different screening courses. Originally three months, the standard course had been shortened over the years and at the time of the study was only six weeks. In addition, due to lack of time and resources, some prescribers had received only an "intensive" four-day or one-week course on prescribing drugs. Some prescribers had received no training at all. Results indicate the longer the period of training, the more likely a prescriber would prescribe fewer drugs. However, while at least six weeks of prescribing training was sufficient to effect a positive change in prescribing habits, a training course of three months was not necessarily better. In facilities where the average number of weeks trained was more than seven (the three-month training course for example), the average number of drugs prescribed dropped only by another 0.04. Furthermore, the statistical significance was weaker.

Perhaps one of the most interesting findings concerned the effectiveness of supervision. While the findings of this study reiterate the findings of previous studies (Guiscafre et al 1988, Guiterrez et al 1994, Agyepong et al 1996, Chowdhury et al 1996) which suggest that supervision is an important component in promoting appropriate drug use, this study revealed a more specific relationship between the type of supervision and prescribing behaviour. On-site supervision and supervision by visiting doctors at least once every two months were found to significantly reduce the average number of drugs prescribed per prescription. However, formal written evaluations conducted once a year, usually by an off-site supervisor, were found to have no significant effect on prescribing. This finding was corroborated by the qualitative data collected in which many prescribers suggested that peer evaluations and supervision from on-site supervisors were more helpful than the yearly formal evaluations. While monthly doctor's visits and on-site supervision were an opportunity to discuss problems regarding prescription, formal evaluations, it was felt, focused less on job performance and more on appearance and conduct. This has important implications for organizations like the CBC that use supervision as a way to promote appropriate drug use.

Other factors that may have contributed to lower drug averages in hospitals and busy clinics were the dental and eye care they provided there. Prescriptions given to patients often contained only one or two drugs.

To reiterate the findings of the prescriber interviews mentioned in 7.1, it is important to note that prescribers felt that drugs were not only curative but also have an important psychological role. Prescribers stated that many patients felt they needed drugs in order to feel better, even if medically the need for medication was not indicated. This perception is similar to findings in Hadiyono et al's 1996 study. Further research at the CBC is needed to establish if this is in fact true or merely a misconception by health facility personnel as was found in Hadiyono et al''s research.

One in three prescriptions filled at CBC facilities contained an antibiotic. The rate was significantly higher in those facilities where prescribers had on average less than six weeks of training on prescribing and where doctors visited less than once every two months.

It appears more difficult to effect a change in antibiotic prescription rates than to reduce the average number of drugs prescribed. While the average drug prescription rate correlated with the extent of formal education, the screening course and mean years of experience prescribing, (i.e. the more education, training and experience the prescriber had, the fewer drugs he/she prescribed), antibiotic prescription rates did not. Furthermore, there was no significant relationship between antibiotic prescription rates and on-site supervision (excluding doctors' visit) or between antibiotic prescription and the formal evaluation.

Quinine and chloroquine prescriptions for the CBC in general were 1:1. However the ratio varied considerably from facility to facility. Reasons for the variation may include the differing prevalence of chloroquine-resistant Plasmodium in each region. In vivo studies completed in Yaoundé (Central Province), Limbe and Douala (South Province) have established chloroquine-resistant Plasmodium in these regions (Brasseur et al 1992). It was therefore expected that the clinics in Kumba, Mutengene and Etoug-Ebe would have lower chloroquine prescription rates.

That quinine rates were comparable to chloroquine rates was a cause of concern for the authors of the standard treatment guidelines, who recommended prescribing Fansidar as the first alternative for chloroquine-resistant Plasmodium. A factor that may have resulted in comparably high quinine prescription rates was the importance CBC personnel placed on considering the economic position of patients. Prescribers stated they would often prescribe the least expensive anti-malarial which is most likely to cure the patient the first time, thereby incurring less cost in the long run. This is corroborated by the finding that from 1996 to 1997 there was a decrease in the chloroquine prescription rate, corresponding to a subsequent increase in quinine prescription rates. However, because the CBC standard treatment guidelines recommend Fansidar as the first line of treatment against chloroquine-resistant malaria, the authors expected to see a corresponding increase in Fansidar, not quinine, when chloroquine prescription rates dropped. However, perhaps reflecting the prescribers' concern for the economic situation of their patients, Fansidar prescription actually decreased from 1996 to 1997. Fansidar is more expensive than quinine. This is substantiated by the results of the case scenarios wherein not a single prescriber considered Fansidar as the first line of treatment in chloroquine-resistant malaria. A detailed discussion of adherence to standard treatment guidelines is discussed below.

Before concluding the discussion regarding the effect that environment and personnel characteristics, training and supervision had on drug indicators, it is equally important to mention those findings which were insignificant. Neither the number of outpatients seen nor number of personnel correlated with either a higher average number of drugs per prescription or with antibiotic rates. Similarly, the duration of consulting time did not correlate with drug indicators examined. The authors expected to see a relationship; a clinic seeing more patients per day would prescribe more drugs per patient in an effort to compensate the patient for waiting longer and spending less time with the consulting nurse. Of the remaining complementary indicators, two previously untested, average cost per prescription and laboratory referral rates did not correlate with average number of drugs prescribed or antibiotic prescription rates. Therefore it was not shown that average cost was a useful indicator of the inappropriateness of drug prescription. Finally, considering the effect the independent variables examined had on the drug indicators, there would have presumably been an effect on injection rates. However, injection rates were impossible to calculate given the quality of the data.